OT, But interesting

[Tim Schmidt]

>From: "Tim Schmidt" <computer_holic@hotmail.com>
>To: linux-user@egr.msu.edu
>Subject: Re: OT, But interesting
>Date: Sun, 06 Aug 2000 17:03:06 GMT
>
>>From: Ben Pfaff <pfaffben@msu.edu>
>>Reply-To: pfaffben@msu.edu
>>To: "Tim Schmidt" <computer_holic@hotmail.com>
>>CC: linux-user@egr.msu.edu
>>Subject: Re: OT, But interesting
>>Date: 06 Aug 2000 10:15:02 -0400
>>
>>"Tim Schmidt" <computer_holic@hotmail.com> writes:
>>
>> > http://www.hardocp.com/news_images/2000/aug2k/080500b.html
>>
>>I'm unable to read that article with lynx, w3m, or links: all I
>>get is links to three "IFRAME"s, which just contain ads, no
>>content.
>
>works in Netscape...
>
>>
>> > It appears that Intel is not only factory overclocking their
>> > CPUs [...]
>>
>>Isn't that an oxymoron, or a misnomer?  As I understand it, when
>>you go to a chip fab, there aren't separate assembly lines
>>cranking out 60 ns and 70 ns, nor are there separate assembly
>>lines cranking out 866 MHz and 933 MHz Pentium IIIs, and so on.
>>Rather, each chip that comes off the line is tested for its
>>performance, where they crank up the speed until it fails at N
>>MHz, and then they sell it as an (N - delta) MHz chip.
>>
>>On the other hand, overclocking is running a chip at a faster
>>speed than recommended by the manufacturer.  So the factory
>>*can't* overclock chips.  They could recommend that you run it
>>faster than it would actually work, I suppose, but in that case
>>it's a defective chip and you should be able to exchange for a
>>working one.
>

I was reading over your question again Ben, and realized an even better way 
to explain why those "questionable tactics" are "factory overclocking."  
Here it goes.  You're right, whe a chip comes off the line, it's tested to 
see how fast it can run -- but the testing is done at default settings (in 
this case, 1.6v, no microcode, stock HSF).  At this point, AMD stamps the 
speed on them and ships them off.  Intel however, takes the fastest of these 
chips, ups the voltage (just as any other OC'er is experienced in doing), 
slaps on a massive HSF, to see how much higher they can get it (in other 
words, it's being overclocked from it's normal tested rating -- at the 
factory).  If it doesn't run 100% stable, they write some new microcode to 
disable whatever part of the processor is giving them problems at that 
speed.  Of course that's an over-simplification of how microcode works, but 
you get the point.

Did this explain it better?  Why it's refered to as factory overclocking?


>Yes Ben, but PIII-C's run at 1.6v (I think, maybe 1.65), Intel has not only
>cranked up the voltage (gets a clearer high/low voltage signal through) to
>increase the speed the chip can reach, they've issued microcode updates for
>the 850 and up which in effect "turn off" parts of the chip that can hold 
>it
>back when reaching higher speeds.  So microcode updates allow the chip to
>clock higher, but have a performance hit.  Also, Intel has outfitted the
>PIII 1000, and 1133 (and I think the 850 and up) chips with massive HSF's
>(Heatsink + fans) to dissipate the massive ammounts of heat produced by the
>high clock-speed/voltage.
>
>In other words, with no tweaking a PIII-C might reach 850Mhz, only with
>--massive-- ammounts of tweaking and/or questionable tactics can a PIII-C 
>go
>any higher.
>
>AMD on the other hand, has not used and microcode updates, voltage hikes,
>die shrinks, or massive heatsinks to get teh Athlon to 1Ghz.  1.1Ghz 
>athlons
>have been in the hands of reviewers for several months now also.
>
>In short, resorting to all these questionable tactics is what is generally
>refered to as "factory overclocking"  -- yes, it is technically an 
>oxymoron.
>
>--Tim
>
>>_______________________________________________
>>linux-user mailing list
>>linux-user@egr.msu.edu
>>http://www.egr.msu.edu/mailman/listinfo/linux-user
>
>________________________________________________________________________
>Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com
>
>_______________________________________________
>linux-user mailing list
>linux-user@egr.msu.edu
>http://www.egr.msu.edu/mailman/listinfo/linux-user

________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com